Project Background Remote sensing image classification using soft computing approach.

2

1.1 Project Background

In general, classification refers to learning classification models or classifiers from data presented by labeled examples Valentina Zharkova, 2007. Image classification is processing techniques which apply quantitative methods to the values in a digital yield or remotely sensed scene to group pixels with similar digital number values into feature classes or categories Purdue University , 2012. Image classification can also called a process which grouped the items objectspatternsimageregionspixels based on the similarity between the item and the description of groups. Common classification procedures can be broken down into two broad subdivisions based on the method used: supervised classification and unsupervised classification. Supervised classification is the process of automatically grouping data into a set of classes by setting up prototypes using a priori knowledge obtained through training. It often involves selection of training data that can represent homogeneous examples of each class Chen, 2010. Unsupervised classification is one of the two basic approaches to digital image classification with the goal of producing land cover maps from remotely sensed data. Unsupervised algorithms evaluate the spectral properties of image pixels and segregate them into naturally occurring statistical groups with little or no guidance from the analyst Rundquist, 2010. The advanced methods that can be used in supervised classification include Parallelepiped Classification Maximum Likelihood Classification, Minimum Distance Classification and Endmember Spectra. As for unsupervised classification, the advanced methods that can be used in unsupervised classification method are K- Means Classification and ISODATA classification. The classifier to be used will be based on the Fuzzy Interference method. The comparison of classifier technique uses between two different types of Unsupervised 3 Technique will be done. In this project, the chosen method is Fuzzy Mamdani and Fuzzy Sugeno for unsupervised image classification.

1.2 Problem Statements